uzbekdev1 / fullstack-webscraper-for-ml

Scrapping relevant info from the web

The documentation in this repository describes the FullStack webscrapping platform for use in Machine learning.

Architecture

We first break the architecture into four distictive components namely Front-End, API, Scrapers and Database. The user sends information from the front-end to the API, the fron-end connects the API through a form. Inputs like the youtube URL are sent through front-end. Later the scrapers through the API pulls the necessary data and is saved to the database. Afterwhich the data is served to the front-end.

The Tech Stack are as below

1. Front-End - javascript
2. API - express
3. scraper - puppeteer
4. db - mysql (typeorm)

Also we need nodejs, npm and mysql.

The Architecture consists of several components:

Front End

For the Front-end we will have a header, an input box and a button. Below which we will have render boxes which renders relevant info from json. This will send data to the API.

API

We will have to create a single route with two methods GET and POST. We use nodejs and simple backed framework express.

Scraper

This function takes in URL and reaches out to YouTube, fetch the relevant data and then store it into the database.

Database

We use mySQL here. Here we add id, name, avatar and channelURL

To run the program

First go into server

$ npm install init

Install all the necessary packages

$ npm install express
$ npm install body-parser

Run the index.js script

$ node index.js

Thanks to Aron from Uber